We worked with 46 8–21 year olds to get their thoughts about the Growing Up on the Online World Consultation that closed on 26 May. Part of that process was to collate their responses, and feed them into the full survey. This is that response.
The full list of questions are here.
Where numbers below are jumped over, it is because we did not have sufficient evidence for reasonable response. “Prefer not to answer” suggests that there was a range of options presented by the young people, making responding to a multiple choice question impossible.
Q9 — What are the benefits of social media use, and being online, for children?
This response is submitted by a researcher working alongside 46 young people aged 8 to 21, exploring the questions asked in the Growing Up in the Online World Young People’s Survey. The findings below reflect what those young people told us, and where our evidence does not support an answer we have said so/will not answer rather than extrapolate beyond the data. Numbers cited throughout the survey reflect the number of participants that provided written evidence for a question; this number differs by topic.
Communication was the most commonly cited written benefit among our participants, by a significant margin (12/17). More spoke of it during discussions in the research session. Participants described social media as essential for arranging in-person meetups, staying in touch with friends and family across distances (in and out of country), and maintaining relationships that would otherwise be difficult to sustain. For young people who are only children, or whose parents work long hours, access to social media was described as a means of social connection where it would not be possible to hang out with friends or be doing other hobbies. Participants who have family in China in particularly noted that a ban for WeChat would see them lose contact; other participants wondered whether messaging tools such as WhatsApp might be included, and how they would then keep in touch with not just their friends, but their parents, during a typical school day.
Participants described social media as a route into communities not available to them locally, including groups organised around shared interests and communities where finding peers with similar experiences face to face was not straightforward. For some young people, online community was filling a gap that their immediate environment had not provided.
Learning and creative development were identified by multiple participants. Three cited YouTube specifically as integral to maintaining or improving off-line hobbies, and one participant described it as necessary for pursuing art-related post-16 ambitions. Two participants noted that their reading and television consumption was directly shaped by social media recommendations. Other participants cited learning as a benefit in more general terms, across various platforms, from Tiktok, to Duolingo, to YouTube and Pinterest. Social media was not described as a replacement for formal education but as a space where skills development and creative interest could be pursued outside school structures.
Social media also functions as a means of decompressing at the end of the school day. Participants described using it after school to relax, and finding such use “relaxing”. This use was described as routine and proportionate rather than compulsive.
Sharing ideas, culture, and news featured across the data, though with lower frequency than communication and learning.
It is worth mentioning that young people are growing up with these technologies and have integrated them into daily life in ways that make the benefits and the tools themselves difficult to separate cleanly. Most participants are fully expecting a ban, and found the lack of specificity in the consultation documentation difficult as a means of conceptualising what they might be discussing. Many are angry that they have been socialised into a system that provides benefits but will be removed from them. The perceived costs of removal were framed not in terms of losing optional extras but in terms of losing infrastructure.
Q10 — What are the harms or risks of social media use, and being online, for children?
Participants found providing specific harm data much harder than specific benefits data. Responses were generic, with no examples of first-hand lived experience. That is not to say that the harms mentioned are not real, or that participants dismissed them. In fact, whilst some participants raised questions around proportionality of action (namely, if you hear about a thing happening on national or international news, that is a sign of an horrific, but outlying, event), many took generic ideas of harm as understood reasons for proposed action.
“Bad influence” and “inappropriate content” were the most commonly cited written concerns, along with “addiction”. “Addiction” was also frequently heard in discussion. It is important to note, however, that when asked about the impact that access to online services had in the participants’ lives, there were no reports of participants struggling to perform their required daily activities (attending school, getting homework done, attending extra-curricular activities) or feeling concern about not having access during, for instance, the school day. The Young Person’s survey asked what participants would do “if they weren’t on social media”, and this generated confusion, because, in a number of cases, it was something that filled 5 minutes of otherwise dead time on the tube, or when waiting to go to school, rather than something consuming significant blocks of time in a way that was concerning to participants. As such, the use of “addicted” as a rhetorical tool, rather than deep-seated concern, must be noted. Within the research sessions, the term functioned as a way of expressing moral ambivalence about their own use (“I know I use it too much”), signalling social norms (“everyone is on it”), or attributing their behaviour to the design of the platform rather than their own choices. The point is being laboured here as the term is perhaps used in a loaded fashion in many policy conversations.
Stalkers and unsafe contact were written by two participants, with “stranger danger” mentioned in the research session by many. One participant referenced a young family member who had been threatened, online and then offline by another member, previously unknown to them, on a shopping platform. The participant — and their family member — seemed to brush off the incident; neither the affected young person nor the family reported the incident to the platform.
Limiting real-world interaction, oversharing, fake news, and eye damage were brought up a limited number of times, again with no specific examples.
Q12 – Would you support a legal requirement for social media services to have a minimum age of access?
MCQ answer: prefer not to answer
Q15 — What do you think the impacts would be of having a minimum age requirement higher than 13 for social media services?
We selected “prefer not to answer” in relation to a minimum age requirement because, although participants did express ideas around minimum ages, this was a cohort that, despite differences in ages, geographical locations and comfort with technology, are expecting to be banned somehow. Stated ages were clearly a proxy: those supporting 13 were broadly happy with the status quo. Those supporting 16 were typically those that used technology less or felt less comfortable with it in general.
What was consistent across all groups, and across the full age range of our participants, was a view that a ban will not work regardless of where the age threshold is set. All participants (even the youngest) were aware of how ineffective the ban in Australia seems to have been.
Participants agreed that as this consultation is currently worded, it is hard and potentially misleading to try and give a single answer for “social media services” as it has not been clearly defined, and the online space is constantly evolving. Participants raised the question of which platforms would be captured.
The impact our participants were most certain about was that a higher minimum age would not stop most young people from accessing social media. Circumvention methods were described across the groups, and the youngest participants had first-hand experience of age verification systems that had assessed them as significantly older than their actual age.
The cliff-edge concern was the most frequently raised argument. Participants across groups described a scenario in which a young person with no prior gradual experience of social media encounters it for the first time at 16, in an uncontrolled and unsupported way. The expectation was that this produces worse outcomes, not better ones. A young person who first encounters social media at 16 without prior gradual experience or education enters a less controlled environment at a later stage of development, without the accumulated experience of navigating it at lower stakes, but at an age where they are less likely to want support from trusted adults like parents or teachers. Participants reflected how they saw older family members struggle with, most recently, spotting AI generated content, precisely as a result of not having had exposure before seeing it presented in an algorithmic feed full of “real” things.
Also recognised by almost all participants: a removal of legitimate access will precipitate a push underground to less safe or scrutinised environments. Discussions were not focused on how specific individuals would get around the ban. It was about what happens to a young person who is using social media in a context where doing so is prohibited: they are less likely to ask for help, less likely to disclose a bad experience, and less likely to involve a trusted adult, because doing so risks exposing behaviour that is technically against the rules.
There is also a democratic element to this. As mentioned already, participants have priced in a ban, and are not happy about it. A number of participants had no idea why the researcher would voluntarily take part in this consultation without being paid. They would never consider doing this themselves; they are already disengaged from the democratic process. It is commonly recognised that political life is in large part currently played out on social media. A ban until 16 years old will see young people be unable to find key political information until, theoretically, the day in which they can vote in a general election. Older participants were puzzled by this.
Within school year groups, a 16 minimum would create inequity for the duration of the academic year. A cohort turns 16 gradually across twelve months. Young people who turn 16 later in the year would face a longer period of restriction than their classmates, creating visible and potentially stigmatising differences within the same year group. This is arguably the case currently, at 13, but those closer to 16 in the participant group clearly felt that this happening at an older age might produce more negative and severe consequences.
The relationship between a 16 minimum and digital literacy education was identified as unresolved. Schools currently face increasing requirements to include aspects of online life as part of PSHE teaching. A ban that removes the practical context in which those lessons might apply makes the education harder to deliver and less grounded in real experience. The connection between formal literacy teaching and actual use is already fragile; further separating them is unlikely to strengthen it.
Q17 — What risks or burdens may be associated with raising the minimum age of digital consent?
Only one group of five participants decided to explore this topic in the research session. In addition to not proving interesting to the majority of participants, it should be noted that the group that did look at it did not fully understand the implications of not being able to consent to personal data being processed at an age older than 13. The researcher had to explain the status quo to them.
Once this was done, the most consistent concern raised was the impact on daily life infrastructure. It is the case that many children feel like they are expected to be outside more; today, this often requires access to digital tools to do so. Participants identified services whose use involves data processing but which are not social media in any meaningful sense: online bus passes, navigation applications such as Citymapper, secure messaging to parents, access to banking apps, streaming services and so on.
Q19 — “There is a case for changing the digital age of consent for some online services but not others”
MCQ answer: prefer not to answer
Q28 — What do you think the impacts would be if online platforms were required to restrict specific features or functionalities, or to introduce time limits?
The preference in our data for daily time limits over overnight restrictions reflects specific concerns about what overnight blocks would mean in practice. Six of eight participants who wrote about this question supported time limits; two supported overnight restrictions. Far more agreed with daily time limits in conversation. The reasons given for not supporting overnight restrictions are worth setting out.
The time-zone concern was raised directly. Young people with friends or family in other countries, specifically the US, described overnight blocks as disproportionately affecting them relative to peers with only local relationships.
Participants raised the question of unusual circumstances: travel, late nights with family, events that fall outside a normal schedule. The objection was not that they were routinely online at 3am but that a blanket restriction would potentially apply on those occasions too, depending upon how it would be implemented. Most nights are ordinary, but not all nights are, and a blanket rule does not distinguish between them.
The framing of the potential restrictions in the consultation text is so broad that many participants discussed fears about being unable to contact parents or similar during an emergency at night. Participants questioned which apps would remain accessible under an overnight restriction and what the position would be if someone needed to contact a parent or trusted adult outside permitted hours. This feels like an over-reaction, but the concern highlights the significant lack of clarity being given.
Parental controls were raised as the more appropriate mechanism. Participants noted that these tools already exist and already allow parents to set time limits and overnight restrictions on specific apps, devices, and even home networks. The concern about platform-level mandates was not that time limits are wrong in principle but that imposing them at the platform level removes the ability of a parent who knows their child to make a different judgment; they are too blunt. Participants also noted, however, that current parental controls are poorly designed and easily circumvented.
Two further observations from the data apply to feature restrictions mentioned in Q24-26 more broadly. Participants noted that persuasive design features, including leaderboards, streaks and limited messaging, are present in educational tools used in schools as well as social media platforms, yet they are noted as being outside of the scope of any proposed restrictions. Participants also observed that these features affect adults as much as children. This was raised not as an argument against child-specific measures in principle but as a proportionality point: where a feature is present across age groups and its effects are not age-specific, child-only restrictions address the symptom in one population without addressing the underlying design incentive.
Q29 — What factors are important when determining which apps, sites or services to apply minimum age of access restrictions to?
Participants struggled to define social media, but broadly agreed on services where you can be a) social with other people and b) share content. They agreed, in theory, that there was a difference between platforms they understood as social media and platforms where social features were present but secondary. What is vital to note, however, is that although this initial definition was agreed on, participants could not then agree on where specific platforms or products fell in the definition. Gaming and music platforms, in particular, could not be easily defined.
The inclusion of “ability to generate non-text mediums such as video or images” as an example in the question illustrates precisely this difficulty. As phrased, this factor either captures almost all social platforms, which allow users to post images and video, or is attempting to identify AI generation tools as a distinct category. If it is the latter, all participants agreed that it points to a class of tool (AI chatbots) that requires separate consideration, since they are not social. We would urge that any factor referencing generative AI be defined with precision and treated separately.
Q31 — What factors are important when determining which apps, sites or services to apply age-restrictions on specific features and functionalities?
Endless scrolling was considered by participants to be the most persuasive design feature. Streaks, autoplay, alerts and notifications, and affirmation features such as likes also featured. Participants noted that these features exist in educational tools as well as social media platforms. They noted that many can already be turned off (and that they do already do this). And they observed that the features affect adults as much as children. Where a feature is present across all ages and its harms are not age-specific, child-only restrictions will not address the underlying incentive structure and may create an uneven playing field between services that comply and those that do not. This raises proportionality questions about child-specific restrictions as the primary mechanism.
Q33 — Are there additional types of service which you think would be appropriate to exempt from age restrictions?
Services that offer effective parental controls as a core feature warrant different treatment. Our research found consistent support for parental oversight and a clear preference for parental controls over platform-level blanket restrictions.
There needs to be a means to protect the use of platforms to further educational interests. This is not just tools used within schools.
Services with limited or incidental social functionality, where user-to-user interaction is not a core feature and is not what draws children to the service, should also be considered for exemption.
Q34 — What are the benefits to children of using AI chatbots?
Our participants were not enthusiastic advocates for these tools. Several noted that AI outputs were often unhelpful, and the prevailing view was that heavy reliance on an AI chatbot as a companion or conversational partner was a sign of worrying loneliness, that would need addressing. Many were confused by the idea of AI companions/romantic partners. The benefits participants identified were practical and specific, not general endorsements of AI chatbot use.
Educational support was the most commonly cited use. Participants described using AI chatbots to help with revision and to practise responses to test questions. The memory function, the ability of a chatbot to retain context across sessions, was specifically noted as useful for learning rather than as a risk feature in this context.
Creative use was identified by younger participants as well as older ones. The value they described was not that AI replaced their own thinking but that it expanded the range of options available to them. It provides a starting point, and provided choices.
School and university integration was raised. Several participants noted that AI tools were increasingly expected in school settings and that AI literacy would be assumed at university level. This was not framed as enthusiasm but as a practical reality, and there was confusion as to why there would be a desire to curtail use.
Use as a search alternative was described too, to explore all manner of things from academic questions to things they did not want to ask a parent.
Q35 — Which AI chatbot features are most risky for children?
MCQ answer: prefer not to answer
Q37 — Should AI chatbots have minimum age restrictions?
MCQ answer: prefer not to answer
Q38 — What do you think the impact would be of introducing age restrictions on AI chatbots or certain features and functions?
Participants showed a low level of AI literacy, on the whole. They were unable to distinguish expected from unexpected chatbot behaviour. As mentioned in Q34, there is a presumption of heavy AI use by individuals in the future. A young person who encounters AI chatbots for the first time at 16 or 18, having had no prior experience or education, is not better placed to navigate these tools safely. As with social media, it is highly unlikely that AI literacy can be effectively taught theoretically.
It remains unclear to the researcher how certain features could be age-restricted. AI chatbots are general-purpose tools. The same interface that supports revision also supports questions about relationships, self-harm, or harmful content. How age restrictions would apply to specific functionalities within a tool of this kind, where the same input field serves all purposes, is not straightforward. Any restriction framework that does not address this architectural reality risks being applied inconsistently, or in ways that produce confusion about what young people are and are not permitted to use across school and home contexts.
Q44 — What methods to circumvent online safety rules do you think children in the UK use, beyond Virtual Private Networks (VPNs), or similar technologies?
Participants of all ages made it clear that circumvention does not require technical sophistication. The methods young people described across all participant groups were exclusively non-technical.
Age verification systems themselves were a recurring source of failure rather than a barrier. Participants across the age range, including the youngest groups, had direct experience of verification systems assessing them as significantly older than their actual age. In one case the estimated age was more than double the participant’s real age. These accounts were consistent and were described with bemusement rather than as deliberate exploitation.
Participants were clear that they should be consulted in relation to this question, although they also recognise that it would be difficult to ask children to respond honestly. In short, though, adults assume that circumvention requires technical effort; children know that – currently at least – it does not.
It should be noted, against the implications of this and the subsequent questions – that many participants in every researh sessions needed VPNs to be explained to them. Participants studying computer science, or coming from countries with more restrictive internet regimes knew what they were. Others did not.
Q52 – Which areas of media or digital literacy do children and families most need additional help with?
MCQ answer:
a. Managing screen time and online habits
b. Spotting adverts, sponsored posts or AI generated content
c. Keeping personal information private
d. Online behaviour and experiences (bullying, respect, comparison or
peer pressure)
e. Checking if information is true
f. Understanding how social media works (for example, ‘likes’ or
algorithms)
g. Staying safe online (including how to have conversations about online
safety)
h. Reporting harmful or upsetting content
i. Knowing which apps or sites are right for their age
Q55 – Outside of schools, how could the UK government better support children and young people to stay safe and feel supported online?
MCQ answer:
a. By providing clear guidance that children can use on their own
b. By supporting parents and carers to support children online
c. By working with online platforms and services that children already
use
d. By supporting youth organisations and community groups to help
children online
e. By making help or advice easy to access when something goes wrong
online
f. By involving children and young people in designing support
Q57 – Who would you trust to determine what is meant by ‘high quality’ online content’ for children 13-16?
MCQ answer:
b. Online platform trust and safety teams
c. Parents, carers or trusted adults
d. Children
e. Developmental experts
g. Youth workers
k. Other (please specify) — Regulators
Q58 – What further action should be prioritised to support positive online spaces for young people?
MCQ answer:
a. Develop best practice principles for industry
b. Develop guidance for parents and carers
c. Develop guidance for children
d. Reviewing international approaches
Q59 — What should be considered when taking further action to support positive online spaces and content for young people?
Participants who engaged with this question noted “positive” felt like a loaded term: what counts as appropriate or high quality content is contested, and different families, communities, and age groups will draw those lines differently.
Among our participants, experts in child development were the unanimous choice as the appropriate authority on what constitutes high quality content for children. Parents and youth workers also featured, although there was wide concern on how different parental stances can be.
The design process also matters. Participants expressed throughout our research that they felt poorly consulted on questions that directly affected them. Further action to support positive online spaces is more likely to be effective and accepted if young people are involved in defining what the goals should be, not only in evaluating the outputs.
Participants were consistent across the research that support for using social media more effectively was preferable to restriction. Further action in this area is more likely to meet young people’s expressed needs if it is oriented around helping them navigate the environment they already use, rather than limiting access to it.
Q60 — To what extent do you agree or disagree with the following statement: “Parents should have control over the online experiences of their children”
MCQ answer: Somewhat agree (option b)
All participants who engaged with this question selected “somewhat agree.”
Parental oversight was broadly accepted as appropriate, and participants described a general preference for parental controls over platform-level blanket restrictions. The reasoning was that parental controls, in principle, allow a more tailored response: a parent who knows their child can make a more calibrated judgment than a platform applying a uniform rule to all users of a given age.
The qualification reflected two consistent concerns. First, parents do not necessarily have better digital literacy than their children. Several participants noted that their parents did not fully understand what they were doing online, and that this limited the usefulness of parental oversight as a safety mechanism. Parental control is only as effective as the parent’s understanding of what they are trying to control. Second, participants were clear that the most effective parental controls were those that came with explanation and discussion rather than being imposed without context. The relationship between parent and child around digital access was described as working best when it was negotiated rather than unilateral.
Q61 — How should this level of control change for children of different ages?
Participants favoured a stepped approach in which parental oversight is progressively reduced as children get older, with a meaningful shift around the age of 16. Parental access to phones and online activity felt as appropriate for younger children.
Around 16, participants identified a shift in expectations. This was not framed as an absolute cliff-edge but as a point at which the expectation of greater privacy becomes reasonable both off- and online, and at which sustained close monitoring starts to feel disproportionate. The general view was that by 16, a young person should be developing their own capacity to manage their online life, supported rather than supervised.
If social media is introduced for the first time at 16, it arrives at the point when parental oversight is already expected to be stepping back. A young person encountering social media for the first time at 16, without prior gradual experience and without close parental oversight, is in a more exposed position than one who has navigated these platforms over several years with active parental involvement. The stepped approach implies that the earlier stages of parental oversight are precisely when the groundwork for safer independent use is laid.
Q62 — What would help parents and carers to more effectively use parental controls?
The most consistent finding on this question is that parental digital literacy is a precondition for effective parental controls, and that improving the tools without improving the literacy is unlikely to produce better outcomes.
Participants described a wide range of parental approaches to digital oversight, from full monitoring of all activity to allowing access to services with minimal involvement.
Current parental controls were described as poorly designed and easily circumvented, including by the youngest participants in our study. The point that emerged most clearly from the data is that parental controls work best when they are accompanied by a conversation between parent and child about what the controls are and why they are in place. Tools imposed without explanation were more likely to be worked around. Tools that reflected a shared understanding between parent and child of what was appropriate and why were more likely to be respected. This suggests that support for parents should focus on building the knowledge and confidence to have those conversations, not only on making the technical tools easier to access.